
Fail Corner: A Decade of Failures

Why OpenAI is merging Codex and ChatGPT and the future of knowledge work | Andrew Ambrosino
Topic 15 of 17
Published Jun 28, 2026 · Clip 59:33–1:01:51 (2:17)
Andrew says this is perhaps the first time he hasn't felt like he was failing. He was a startup founder for a long time, selling the company 'for parts' after years of slog in heavily regulated spaces; a later AI-tools startup in another locked-down industry felt like time after time of things not working. He estimates he failed for somewhere between 10 and 15 years before Codex and is still surprised daily that things go well. Even now there are constant micro-failures: proposed product shapes get posted in Slack and met with 2,000-message threads about how stupid they are — which he loves, because that brutal internal honesty is why the external product has been great.